Network virtualization and multi-tenancy

So far, we've talked about several hypothetical situations where you may have multiple segregated departments or external customers deploying and managing multiple resources inside of SCVMM. We've covered separating out individual access to resources using role groups, and from a compute perspective, the virtualization layer does a very good job of isolating virtual workloads from VM to VM, but what about virtual networking? If these entities are all utilizing the same hypervisors, all virtual machine guest traffic is most likely going out across shared links, right? How does VMM help us separate out that traffic?

Historically, this has been a function that has been taken care of by creating dedicated VLANs for each individual entity. VLANs are still a viable option in today's world. However, the utilization of VLANs can become problematic when scaled out beyond a certain point, and it requires quite a bit of management from some sort of network operations...